The computing power required to model brain subsystems at any level of detail is enormous. Compartmental models of individual neurons, which involve solving sets of differential equations for each compartment make very heavy use of processor time. Network models of even a small region of brain must include large numbers of synapses which make very heavy demands on memory usage. And cell models are likely to become more detailed and biologically realistic as more experimental details of their internal dynamics are discovered.
This suggests the need for parallel or distributed computers to allow larger or more detailed models to be investigated than can fit on a single workstation. This was the motivation for the development of PGENESIS, the parallel version of Genesis, and large models have been run on supercomputers using this tool. The problem with hand coding a parallel simulation using C++ or PGENESIS is the amount of development and debugging time needed to get a parallel program running efficiently, time which would be better spent on modelling issues.
An aim of Neosim is to allow small models developed on workstations to be scaled up and run on networks of workstations and supercomputers without having to write explicitly parallel code.